OpenAI launches Daybreak to take on Anthropic’s Mythos in cyber defence

The new platform pairs GPT-5.5 variants with Codex Security and a roster of enterprise security partners, all aimed at defenders.

OpenAI has unveiled Daybreak, a cybersecurity initiative aimed at finding software vulnerabilities, generating patches, and validating fixes inside enterprise codebases. The launch positions OpenAI directly against Anthropic’s Mythos, which has spent the past few months dominating the conversation about AI-powered defence.

Daybreak rests on three model variants, according to OpenAI. GPT-5.5 covers general-purpose use under standard safeguards. GPT-5.5 with Trusted Access for Cyber is reserved for verified defenders performing tasks such as secure code review, vulnerability triage, malware analysis, and patch validation.

A third model, GPT-5.5-Cyber, is a more permissive variant for authorised red teaming, penetration testing, and controlled validation.

The platform’s working method begins with threat modelling against a given repository, then identifies and tests vulnerabilities in an isolated environment, and finally proposes ...